Heh I jacked my settings to auto reject on 2 flags and moderate on 1. Been making a huge difference ever since the switch to 1.4. Being a gaming forum I don't think we will have more than a couple of false positives a year on 2 flags. However the one that annoys me is single flag on username generates plenty of false positives for the moderation queue.
At this point we get the exact same number of spammers with or without the captcha so its like... Why bother having it at all. The honeypot add-on in 1.3 was brilliant. No captcha. No spammers. Either human spammers have increased or people are wrong but right after switching to 1.4 we ended up with a group of very mechanical very predictable spammers. Females with 2 names, often in their 20s. Pakistani, Indian or data center IPS. They look at their own profile after registering.
Since we are a male dominant forum almost tempted to write a plugin to auto moderate females.